Skip to main content

Redistricting ensembles

Project description

rdaensemble

Redistricting ensembles

Methods

The code in this repository supports several methods for generating ensembles of redistricting plans (maps):

  • Random maps from random spanning trees (RMfRST)
  • Random maps from random starting points (RMfRSP)
  • Ensemble of maps using MCMC/ReCom (ReCom)
  • Ensemble of maps using Sequential Monte Carlo (SMC) <<< TODO

Input Files

The inputs for generating & scoring ensembles are:

from rdascore import load_data, load_shapes, load_graph, load_metadata

data: Dict[str, Dict[str, int | str]] = load_data(data_path)
shapes: Dict[str, Any] = load_shapes(shapes_path)
graph: Dict[str, List[str]] = load_graph(graph_path)
metadata: Dict[str, Any] = load_metadata(state_code, data_path)

The precinct data, shapes, and graphs are all available in the companion repository rdatools/rdabase in the data directory by state. They are named NC_2020_data.csv, NC_2020_shapes_simplified.json, and NC_2020_graph.json, for example.

Theoretically, these inputs can come from any source, but for simplicity, reproducibility, and apples-to-apples comparisons, it's best to use the input files in rdabase.

Output Files

Ensembles are saved as JSON files. A file contains metadata about the ensemble, including the method used to generate it, and then a plans key with a list of plans:

plans: List[Dict[str, str | float | Dict[str, int | str]]]

Each plan item has a name (str), an optional weight (float), and a plan (Dict[str, int | str]]) which represents the assignments as geoid: district_id key: value pairs.

Scores for the plans in an ensemble are saved as a CSV file, with one row per plan and one column per metric. The metrics are the same as those produced by rdatools/rdascore, except they also include the energy of the plan. The metric names are descriptive.

When a scores CSV file is produced, a companion JSON file with metadata about the scoring is also generated.

Naming Conventions

You can name ensemble and score files anything you want. To facilitate understanding the contents of these files without having to open them, we recommend the following the convention:

  • Ensemble example: NC20C_RMfRST_1000_plans.json
  • Scores example: NC20C_RMfRST_1000_scores.csv

where "NC" is the state code, "20" stands for the 2020 census cycle, "C" abbreviates "Congress" (as opposed to state upper or lower house), "RMfRST" is the method, 1000 is the number of plans in the ensemble, and "plans" and "scores" distinguish between the two types of files.

Note: The scores metadata file will be named the same as the scores file, except it will end _metadata.json instead of .csv, for example, NC20C_RMfRST_1000_scores_metadata.json.

Usage

To generate an ensemble of 1,000 plans using the random maps from random spanning trees method (RMfRST), run:

scripts/rmfrst_ensemble.py \
--state NC \
--data ../rdabase/data/NC/NC_2020_data.csv \
--shapes ../rdabase/data/NC/NC_2020_shapes_simplified.json \
--graph ../rdabase/data/NC/NC_2020_graph.json \
--size 1000 \
--plans ~/iCloud/fileout/ensembles/NC20C_RMfRST_1000_plans.json \
--log ~/iCloud/fileout/ensembles/NC20C_RMfRST_1000_log.txt \
--no-debug

To score the resulting ensemble, run:

scripts/score_ensemble.py \
--state NC \
--plans ~/iCloud/fileout/ensembles/NC20C_RMfRST_1000_plans.json \
--data ../rdabase/data/NC/NC_2020_data.csv \
--shapes ../rdabase/data/NC/NC_2020_shapes_simplified.json \
--graph ../rdabase/data/NC/NC_2020_graph.json \
--scores ~/iCloud/fileout/ensembles/NC20C_RMfRST_1000_scores.csv \
--no-debug

To generate random maps from random starting points (RMfRSP) instead, use the rmfrst_ensemble.py script. For ReCom, use the recom_ensemble.py script.

Note: Ensemble JSON files can be quite large, bigger than GitHub's 100 MB file size limit, so we recommend that you write them to the ensembles directory, which is ignored by Git.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

rdaensemble-1.1.5.tar.gz (15.5 kB view details)

Uploaded Source

Built Distribution

rdaensemble-1.1.5-py3-none-any.whl (18.7 kB view details)

Uploaded Python 3

File details

Details for the file rdaensemble-1.1.5.tar.gz.

File metadata

  • Download URL: rdaensemble-1.1.5.tar.gz
  • Upload date:
  • Size: 15.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.2 CPython/3.10.6

File hashes

Hashes for rdaensemble-1.1.5.tar.gz
Algorithm Hash digest
SHA256 7d4fdde8dac814ffd32f5a8cc6642037413e75c607170a236a9475be249abbcd
MD5 cc34f99c8127bbed2027295308c8a086
BLAKE2b-256 890d45bd72f905aacf0f0a0f4bebd726098e4be24fc043c82a8e8f8d462a3d6a

See more details on using hashes here.

File details

Details for the file rdaensemble-1.1.5-py3-none-any.whl.

File metadata

  • Download URL: rdaensemble-1.1.5-py3-none-any.whl
  • Upload date:
  • Size: 18.7 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.2 CPython/3.10.6

File hashes

Hashes for rdaensemble-1.1.5-py3-none-any.whl
Algorithm Hash digest
SHA256 b42bb84c748bdf32343d54e3783f43e86cdf5ebd8e34f0f30120680f5190011d
MD5 61029b820f05b0b3315413404e6d92f9
BLAKE2b-256 d9dff39e48425971daf6a884c3289104a242eb4d56152a9a5fff4400b1bfcc5a

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page